magic-timer

magic-timer

pip install magic-timer

Conveniently get a rough idea of how long things take.

This is a light wrapper around the standard library's time.monotonic. It aims to provide a clean API, and nice output strings.

How to use:

from magic_timer import MagicTimer

timer = MagicTimer(history=True)
for i in range(3):
    expensive_computation()
    # Print nicely formatted string:
    print(f"{i} - elapsed time {timer}")

# Get the elapsed times that were printed:
print("timer.str_history =", timer.str_history)

0 - elapsed time 510 milliseconds
1 - elapsed time 1.1 seconds
2 - elapsed time 1.6 seconds
timer.str_history = [0.5046274580000158, 1.005028416000016, 1.510260250000016]

Use via context manager:

from magic_timer import MagicTimer

with MagicTimer() as timer:
    # do stuff
    x = sum(i*i for i in range(100_000))

# Print a nicely formatted string:
print('Stuff took', timer)

# Or get the elapsed time in seconds:
time_elapsed = timer.time_elapsed()
print(time_elapsed)

Stuff took 8.0 milliseconds
0.007906290999997623

Use via MagicTimer object:

from magic_timer import MagicTimer

def do_stuff():
    [i*i for i in range(5_000_000)]

timer = MagicTimer()
do_stuff()
print('Stuff took', timer)
do_stuff()
print("Note the timer's still ticking, unless `.stop()` is called...", timer)

Stuff took 210 milliseconds
Note the timer's still ticking, unless `.stop()` is called... 400 milliseconds

To pause the timer, use the stop method (restart with the .start() method). (Note that the context manager automatically calls .stop()).

from magic_timer import MagicTimer

def do_stuff():
    [i*i for i in range(5_000_000)]

timer = MagicTimer()
do_stuff()
timer.stop()
print('Stuff took', timer)
time_elapsed = timer.time_elapsed()
other_stuff()
timer.start()  # continue timing

Use via ftimer decorator:

from magic_timer import ftimer

@ftimer
def do_stuff():
    [i*i for i in range(20_000_000)]

do_stuff()

`do_stuff` ran in 1.9 seconds.

The use case for this package:

You have something you want to time, but you don't want to time it multiple times with timeit.

You also don't want to use Jupyter's %%timeit because it puts the cell into a different scope.

You can import magic-timer, throw it in, and get a rough idea of the time taken. (It's slightly neater than using time.monotonic directly.)